ABSTRACT
This chapter presents a new algebraically derived feature for geometric pattern recognition whose performance is better than that of the schemes. one is interested in recognizing objects irrespective of their position, orientation, and scaling. Numerous geometric pattern recognition algorithms have been proposed in the literature, and some of these have been based on mathematical morphology. The chapter presents a few morphologically derived features for complete as well as pseudocharacterization of multidimensional objects irrespective of their position, orientation, and size. All the algorithms for pseudocharacterization lend themselves to parallel implementation. New algebraically derived features of images were proposed in an attempt to recognize a two-dimensional binary image directly in its discrete domain. These features were defined using morphological operations. In the area of computational geometry, several algorithms have been proposed for exact as well as approximate congruences. Even the most general form of this scheme is unable to classify objects uniquely, and, moreover, it is computationally intensive.
